The paper describes the development and the experimental validation of a cryogenic magnetic shielding system for transition edge sensor based space detector arrays. The system consists of an outer mu-metal shield and an inner superconducting niobium shield. First, a basic comparison is made between thin-walled mu-metal and superconducting shields, giving an off-axis expression for the field inside a cup-shaped superconductor as a function of the transverse external field. Starting from these preliminary analytical considerations, the design of an adequate and realistic shielding configuration for future space flight applications (either X-IFU [D. Barret et al., e-print arXiv:1308.6784 [astro-ph.IM] (2013)] or SAFARI [B. Jackson et al., IEEE Trans. Terahertz Sci. Technol. 2, 12 (2012)]) is described in more detail. The numerical design and verification tools (static and dynamic finite element method (FEM) models) are discussed together with their required input, i.e., the magnetic-field dependent permeability data. Next, the actual manufacturing of the shields is described, including a method to create a superconducting joint between the two superconducting shield elements that avoid flux penetration through the seam. The final part of the paper presents the experimental verification of the model predictions and the validation of the shield's performance. The shields were cooled through the superconducting transition temperature of niobium in zero applied magnetic field (<10 nT) or in a DC field with magnitude ∼100 μT, applied either along the system's symmetry axis or perpendicular to it. After cool-down, DC trapped flux profiles were measured along the shield axis with a flux-gate magnetometer and the attenuation of externally applied AC fields (100 μT, 0.1 Hz, both axial and transverse) was verified along this axis with superconducting quantum interference device magnetometers. The system's measured on-axis shielding factor is greater than 106, well exceeding the requirement of the envisaged missions. Following field-cooling in an axial field of 85 μT, the residual internal DC field normal to the detector plane is less than 1 μT. The trapped field patterns are compared to the predictions of the dynamic FEM model, which describes them well in the region where the internal field exceeds 6 μT.